In the age of AI, the goals of observability extend far beyond mere system stability and high performance. Given the inherent uncertainty of Large Language Models (LLMs) and AI Agents, detailed tracking, evaluation, and optimization have become more critical than ever.

Litefuse is an Agent observability and evaluation platform built on Apache Doris. It supports evaluation-driven development, helping developers build and deliver more reliable AI Agents.

Presentation Highlights:

1. The importance of Agent Observability and Evaluation.
2. Key challenges in Agent Observability and Evaluation.
3. Building Agent Observability and Evaluation based on Apache Doris.

Speakers:

Kang Xiao: SelectDB, Co-Founder

SelectDB, Co-Founder 360 and QianXin, Director Baidu, Senior Engineer